1. Field of the Invention
The present invention relates to an image processing technique of detecting and managing the face region of an object in image data.
2. Description of the Related Art
As digital still cameras (to be also referred to as “DSCs” hereinafter) become popular, image data that numbers as many as several thousands or several tens of thousands need to be handled nowadays. As one especially important technique for the handling method, personal recognition is implemented by handling images based on a person's face. For example, a face region included in an input image is detected in advance, information (to be referred to as a “feature amount” hereinafter) obtained by analyzing the detected face image is extracted, and the extracted feature amount is registered. Note that the feature amount to be registered increases in accordance with the number of images to be analyzed and the number of faces included in an image. A database in which a plurality of feature amounts are registered will be called a “dictionary” or “face dictionary”. Personal recognition becomes possible by collating an obtained face dictionary with a feature amount obtained by analyzing a newly input image.
Japanese Patent Laid-Open No. 11-203415 has proposed a method of categorizing a feature amount in a dictionary. First, learning samples of patterns necessary to create an identification dictionary are prepared in advance, and similar patterns are categorized. Then, the category is subdivided to reduce a recognition error within the created category of similar patterns. After the subdivision, subdivided similar categories are integrated. Based on the categorization result, a feature amount is registered in the dictionary.
However, in the similar pattern category identification dictionary creation apparatus disclosed in Japanese Patent Laid-Open No. 11-203415, similar patterns are not categorized for each object, and the recognition accuracy may become poor depending on an object.